loop-skills AGENTS.md

AGENTS.md

Guidelines for AI agents working in this repository.

Repository Overview

This repository contains Agent Skills following the Agent Skills specification, packaged as a Claude Code plugin marketplace via .claude-plugin/marketplace.json.

• Name: loop-skills
• GitHub: donnfelker/loop-skills
• License: MIT

Every plugin here is a bounded autonomous loop: it drives an artifact to a terminal state, checks an explicit termination condition (quiet round, two clean passes, a cap), and commits per round for an audit trail.

Repository Structure

plugins/
└── plugin-name/
    ├── .claude-plugin/
    │   └── plugin.json        # Plugin manifest (name, description, version, license)
    ├── references/            # Optional - plugin-wide docs cited via ${CLAUDE_PLUGIN_ROOT}/references/...
    └── skills/
        └── skill-name/
            ├── SKILL.md       # Required skill file (<500 lines)
            ├── references/    # Optional - skill-local docs loaded on demand
            ├── scripts/       # Optional - executable code
            └── assets/        # Optional - templates, data files

Build / Lint / Test Commands

Skills are content-only (no build step). Validate with:

./validate-skills.sh plugins/*/skills/

Manually verify: YAML frontmatter is valid; name matches the directory exactly; name is 1-64 chars, lowercase alphanumeric and hyphens; description is 1-1024 characters.

Required Frontmatter

---
name: skill-name
description: What this skill does and when to use it. Include trigger phrases.
---

Field	Required	Constraints
name	Yes	1-64 chars, lowercase a-z, numbers, hyphens. Must match dir.
description	Yes	1-1024 chars. Describe what it does and when to use it.
license	No	License name (default: MIT)
metadata	No	Key-value pairs (author, version, etc.)